Principal Software Engineer - AI

New Balance•Boston, MA

1d•Hybrid

About The Position

New Balance is seeking an experienced Principal Software Engineer to join our Enterprise AI Solutions team. This team is responsible for researching and delivering AI based solutions across the company to improve internal workflows, decision-making, and operational efficiency across a wide range of business functions, including operations, supply chain, finance, product, and corporate services. A core focus of this role is designing and building enterprise-grade AI assistants and automations using RAG and agentic AI techniques —software agents that can plan, coordinate tools, interact with systems, and execute multi-step workflows with appropriate guardrails. These approaches are becoming an important part of modern enterprise software architectures, particularly for scaling automation responsibly across complex organizations. In this role, you will work closely with cross-functional partners, data and platform teams, and other engineers to design, implement, and deploy AI-enabled systems that are reliable, maintainable, and aligned with business needs. This position is based in our Brighton, MA corporate office with occasional travel to Lawrence, MA.

Requirements

10+ years of experience in software development , with at least 4 years focused on AI/Machine Learning and Large Language Models.
Proven Experience in Agentic Frameworks: Hands-on experience building with frameworks such as LangChain, LangGraph, or Semantic Kernel.
Mastery of the Microsoft AI Ecosystem: Experience with Azure OpenAI Service, Azure AI Search, Microsoft Copilot Studio and related areas
Advanced Python: Expert-level Python skills, with a focus on building scalable, highly concurrent, maintainable enterprise backends.
Comfortable with Front End development using React or similar
Systems Integration: Strong background in designing RESTful APIs and integrating AI agents with legacy and modern enterprise systems (ERPs, CRMs, and SQL/NoSQL databases).
Architectural Vision: Ability to distinguish between "hype" and "utility," choosing the right AI approach (RAG vs. Fine-tuning vs. Prompt Engineering) for the specific business problem.
Effective Communicator: Ability to explain complex AI logic and agentic workflows to non-technical business stakeholders using clear, visual communication.
Education: Degree in Computer Science, Data Science, or a closely related field.

Nice To Haves

Vector Database Proficiency: Experience with specialized vector stores such as Azure AI Search.
Agentic Memory & State: Familiarity with implementing persistent memory and long-term state management for multi-session AI agents.
DevOps for AI: Experience setting up LLMOps pipelines for monitoring model performance, drift, and hallucination rates.
Frontend Basics: Familiarity with Chainlit or Streamlit for rapid internal prototyping of agent interfaces.
Security Background: Understanding of prompt injection mitigation and secure data handling in GenAI applications.

Responsibilities

Define the architectural blueprint for autonomous and semi-autonomous agents that execute multi-step business processes across disparate enterprise systems.
Partner with leadership in Finance, Supply Chain, and Product to identify high-value automation opportunities and translate them into scalable AI solutions.
Lead the selection and implementation of models, tools, frameworks, and SDKs across Open Source and Microsoft AI ecosystem (Microsoft Foundry and Copilot Studio) ensuring all solutions meet enterprise standards for security, privacy, and cost-efficiency.
Design robust "human-in-the-loop" systems and technical guardrails to ensure AI agents operate reliably and safely within a complex corporate environment.
Oversee the full lifecycle of AI deployments—from initial Proof of Concept (PoC) to production-grade integration with internal APIs and databases.
Act as a subject matter expert on Large Language Model (LLM) orchestration, guiding other engineers on best practices for RAG (Retrieval-Augmented Generation) and agentic reasoning.